Modeling and prediction of criteria pollutants over the urban areas is essential for the formulation and improvisation of urban air quality management strategies. Various statistical techniques have been employed worldwide for accurate prediction of the air pollutants. This study focuses on the analysis and prediction of the criteria pollutants over a tropical urban area (Durgapur, 23̊ 30′ 34.58″ N and 87̊ 21′ 03.42″ E) performed by using statistical models viz. multiple linear regression (MLR) and principal component regression (PCR). Multiple linear regression analyses have been performed using the original variables and principal components (PCs) as the inputs. On the basis of the performance indicators, MLR model is found to perform better than the PCR in most cases. The R² values obtained by MLR are 0.962, 0.945, 0.898, 0.937, 0.603, 0.874, 0.871, 0.837, 0.858, 0.868, 0.842 and 0.825 for PM₁₀, PM_2.5, sulphur dioxide, nitrogen dioxide, carbon monoxide, ammonia, ozone, benzene, benz(a)pyrene, arsenic, lead and nickel respectively which are greater than the respective R² values obtained by PCR model. Results of the two models reveal that use of PCA could not enhance the MLR performance. The predictive equations proposed by the statistical models suggest that the meteorological parameters (temperature, relative humidity, wind speed and cloud cover) have significant influence on the concentration of the criteria pollutants.